Our objective was to investigate whether steadily increasing resistance levels are inevitable in the course of a failing but unchanged Highly Active Antiretroviral Therapy (HAART) regimen. Patients having an unchanged HAART regimen and a good CD4 response (100 cells/microl above nadir) despite consistent HIV-RNA levels above 200 copies/ml were included in the study. The study period spanned at least 12 months and included 47 plasma samples from 17 patients that were sequenced and analysed with respect to evolutionary changes. At inclusion, the median CD4 count was 300 cells/ml (inter-quartile range (IQR): 231-380) and the median HIV-RNA was 2000 copies/ml (IQR: 1301-6090). Reverse transcription inhibitor (RTI) mutations increased 0.5 mutations per y (STD = 0.8 mutations per y), while major protease inhibitor (PI) resistance mutations increased at a rate of 0.2 mutations per y (STD = 0.8 mutations per y) and minor PI resistance mutations increased at a rate of 0.3 mutations per y (STD = 0.7 mutations per y). The rate at which RTI mutations accumulated decreased during the study period (p = 0.035). Interestingly, the rate of mutation accumulation was not associated with HIV-RNA level. The majority of patients kept accumulating new resistance mutations. However, 3 out of 17 patients with viral failure were caught in an apparent mutational deadlock, thus the development of additional resistance during a failing HAART is not inevitable. We hypothesize that certain patterns of mutations can cause a mutational deadlock where the evolutionary benefit of further resistance mutation is limited if the patient is kept on a stable HAART regimen.